GPU Computing

[Yukterez]

If I install a budget GPU which has about the same power (PassMark) as the integrated graphics of the CPU (in this case i5-8400 + UHD 630), will it free up some CPU ressources and system RAM, or would it just be a total waste which will make my system slower? Just to be clear, I don't want to use it for gaming, but for CUDA/OpenCL-related programming and maybe some video rendering. Most high priced GPUs seem optimized for gaming, so I wouldn't want to spend extra cash for something I don't need...

[TheGlenlivet]

12 minutes ago, Yukterez said:

If I install a budget GPU which has about the same power (PassMark) as the integrated graphics of the CPU (in this case i5-8400 + UHD 630), will it free up some CPU ressources and system RAM, or would it just be a total waste which will make my system slower? Just to be clear, I don't want to use it for gaming, but for CUDA/OpenCL-related programming and maybe some video rendering.

If you use a program that allows GPU assisted rendering then you will see a BIG jump in performance, not a reduction.  What Software do you use?

[Yukterez]

Mostly Mathematica and sometimes Matlab, which both support CUDA/OpenCL. My project at the time is relativistic raytracing which I would like to optimize for faster rendering. I'm asking because this guy is warning about making the system slower if installing a card which is not stronger than the onboard video.